Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1 23
Social Psychiatry and PsychiatricEpidemiologyThe International Journal for Researchin Social and Genetic Epidemiology andMental Health Services ISSN 0933-7954 Soc Psychiatry Psychiatr EpidemiolDOI 10.1007/s00127-014-0932-y
The multilevel determinants of workers’mental health: results from the SALVEOstudy
Alain Marchand, Pierre Durand, VictorHaines & Steve Harvey
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag Berlin Heidelberg. This e-offprint is
for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.
ORIGINAL PAPER
The multilevel determinants of workers’ mental health: resultsfrom the SALVEO study
Alain Marchand • Pierre Durand • Victor Haines III •
Steve Harvey
Received: 24 February 2014 / Accepted: 16 July 2014
� Springer-Verlag Berlin Heidelberg 2014
Abstract
Purpose This study examined the contribution of work,
non-work and individual factors on workers’ symptoms of
psychological distress, depression and emotional exhaus-
tion based on the multilevel determinants of workers’
mental health model.
Methods Data from the SALVEO Study were collected in
2009–2012 from a sample of 1,954 employees nested in 63
workplaces in the province of Quebec (Canada). Multilevel
regression models were used to analyse the data.
Results Altogether, variables explain 32.2 % of psycho-
logical distress, 48.4 % of depression and 48.8 % of
emotional exhaustion. Mental health outcomes varied
slightly between workplaces and skill utilisation, physical
and psychological demands, abusive supervision, inter-
personal conflicts and job insecurity are related to the
outcomes. Living in couple, having young children at
home, family-to-work conflict, work-to-family conflict,
strained marital and parental relations, and social support
outside the workplace associated with the outcomes. Most
of the individual characteristics also correlated with the
three outcomes. Importantly, non-work and individual
factors modulated the number and type of work factors
related to the three outcomes.
Conclusion The results of this study suggest expanding
perspectives on occupational mental health that fully rec-
ognise the complexity of workers’ mental health
determinants.
Keywords Mental health � Work conditions � Family �Social network � Individual characteristics
Introduction
Despite the large number of studies carried out over the last
two decades, there are still important debates about the
causes of mental health problems in the workforce [10, 16].
Many studies have pointed out the role of occupations and
work conditions related to decision latitude, demands,
social support and rewards. These conditions are at the very
heart of the dominant demands-control [39], demands-
control-support [40], effort-reward imbalance [66], and the
job demands-resources [26] theoretical frameworks. How-
ever, studies on the etiology of mental health problems
underscore the non-work and individual factors which are
often neglected in work-stress studies [10, 16]. Also, few
studies examine the relative contribution of work condi-
tions when non-work and individual characteristics are
accounted for, and few research design conduct joint
analyses of workers’ work and non-work experiences that
would allow us to disentangle how these many conditions
contribute to psychological distress, depression and burn-
out simultaneously.
The aim of the study is to analyse the contribution of the
multilevel determinants of workers’ mental model that
integrates work, non-work and individual factors to the
A. Marchand (&) � P. Durand � V. Haines III
School of Industrial Relations, University of Montreal, CP 6128,
Succursale Centre-ville, Montreal, QC H3C 3J7, Canada
e-mail: [email protected]
P. Durand
e-mail: [email protected]
V. Haines III
e-mail: [email protected]
S. Harvey
John Molson School of Business, Concordia University, 1455 De
Maisonneuve Blvd. West, Montreal, QC H3G 1M8, Canada
e-mail: [email protected]
123
Soc Psychiatry Psychiatr Epidemiol
DOI 10.1007/s00127-014-0932-y
Author's personal copy
examination of symptoms of psychological distress,
depression and the emotional exhaustion component of the
burnout process.
The multilevel determinants of workers’ mental health
The multilevel determinants of workers’ mental health are
modelled here in response to previous work-stress models
that have failed to integrate, theoretically and empirically,
components of the social environment (i.e., workplace,
family, social network) in which are embedded workers.
The integrative model we advance views workers within
their broader social environment composed of various
structures with which they interact in their daily lives.
These interactions can be sources of pleasure and well-
being, but they can also be sources of suffering that may
affect the psychic balance. Micro–macro and agency-
structure sociological approaches define social structures
and agent personality as conditions of social action that
determine a set of constraints and resources that shape
individuals’ contingencies, locations, and opportunities [6,
7, 19, 29, 69]. At the macro level, macrosocial structures
represent social arrangements tied to the economic, polit-
ical and cultural systems, as well as to the system of
stratification, diversification, and social integration of a
society at the national level. At the meso level, structures
of daily life constitute intermediate arrangements between
individuals and macrosocial structures that organise the
basis of everyday life, routines, and affective ties. These
include the workplace, the family, and the social network.
At the micro level, agent personality represents the con-
straints and resources associated with reflexivity, rational-
ity, creativity, demography, affect, the body, biology,
representations, perceptions, motivations, habits, and atti-
tudes [6, 19, 29, 69]. Agent personality is thus not limited
to psychological traits or personality structure as under-
stood in psychology. It is an overall representation of
individual-specific conditions constructed around the body,
the mind, and the social environment. The relationships
between the agent and the social structures are dialectical
as they produce reciprocity and interaction as action
modalities that may result in unintended consequences
unanticipated by agents [7, 29]. Workers’ mental health
problems can thus be viewed as unintended consequences
of action being influenced simultaneously by social struc-
tures and agents’ constraints-resources. Constraints are
stressors that have the potential to affect an individual’s
adaptability [58] and to lead to potential imbalances in
one’s physiological and mental systems [24]. Resources are
protective factors against environmental stressors, but they
are not necessarily effective for everyone [57].
In this study, structures of daily life and agent person-
ality are at the core of our analysis. As far as work is
concerned, we may expect psychological distress, depres-
sion, and emotional exhaustion to vary according to
workers’ position and experiences in their respective firms,
and across firms. Firms distinguish themselves around
several characteristics related to their environment, eco-
nomic sector, profitability, organisational culture, human-
resources practices, occupational health and safety pro-
gramme, and so forth. Therefore, the general context of the
firm can be a source of variance in symptoms of mental
health experienced by workers. Indeed, a recent study
found that workers’ cortisol secretions, which is a hormone
associated with the physiological stress response of an
individual and correlated with mental health symptoms
[21], varied across workplaces even after adjusting for
work hours, gender, age and work–non-work days [50].
Work organisation conditions experienced by workers
within a firm can also contribute to mental health. They
relate to task design, demands, social relations, and grati-
fications [48]. Task design is about the level of skill util-
isation and decision authority workers are allowed when
performing tasks. Better mental health status has been
associated with higher levels of skill utilisation and deci-
sion authority [37, 62, 72]. Likewise, poorer mental health
has been associated with higher work demands, including
when these come in the form of physical demands (envi-
ronment and individual efforts) [15, 49, 83], psychological
demands (work pace, quantity of work, conflicts) [37, 59,
72], and contractual demands (number of working hours,
irregular work schedule) [4, 27, 68]. As for social relations
with colleagues and managers, when workers are supported
at work it is known to be associated with fewer mental
health problems [15, 43, 72]. However, abusive supervision
[31, 52, 76] and social relationships involving aggression
and violence at work frequently relate to poor mental
health status [34, 35, 49, 55]. With respect to gratification
from work, past research has shown that expectation of job
recognition and career perspectives are related to better
mental health [2], while job insecurity fosters decreases in
mental health status [15, 18, 74].
Consideration of family and social network outside the
workplace defines constraints and resources in terms of
marital and parental status, strained marital and parental
relations, levels of household income, and social support
from one’s social network outside the workplace. Fewer
mental health problems are expected when living with a
partner [4, 47, 80], in households with young children [42,
47, 79], and those that have low-strain relationships with
spouse or children [8, 23, 47], higher household incomes
[41, 84], less work-family conflicts [12, 33, 75], and greater
access to the support of social network outside the work-
place [22, 47, 73].
Characteristics of agent personality include gender, age,
physical health, psychological traits (self-esteem, locus of
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
control), lifestyle habits (alcohol intake, smoking, physical
activity), and stressful childhood events. Mental health
problems are more prevalent for women [1, 47, 80], less
with increasing age [15, 17, 47], and more prevalent when
physical health problems are reported [47, 78, 86]. Higher
self-esteem and an internal locus of control should lessen
mental health problems [28, 47, 71]. Life habits that
involve high levels of alcohol intake [5, 47, 85], tobacco
use [46, 47, 85], and less physical activity [45, 56] all seem
to increase the likelihood of mental problems. Finally,
mental health problems may be more pronounced when
stressful childhood events such as a parental death, divorce,
and alcohol or drug problems in the family were experi-
enced [47, 81].
In summary, the multilevel determinants of workers’
mental model views work factors as one possible mecha-
nism explaining workers’ mental health considering other
structures of daily life in which individuals are embedded
as well as their individual characteristics. Therefore, ana-
lysis of workers’ mental health must integrate the role of
family situation, social network and characteristics of agent
personality in order to examine the specific contribution of
workplace stressors.
Materials and methods
Data
The SALVEO study was conducted in Canada and aimed
to evaluate the contribution of work, family, individual
characteristics and social network to workers’ experience
of mental health problems. Data were collected in
2009–2012 within 63 Canadian workplaces, randomly
selected from a list of client companies of a large insurance
company. These companies were invited by their insurer to
participate in this study and those accepting the invitation
were referred to the research team. At this stage, the
response rate was 41.0 %, which is significantly higher
than the ones usually found in organisational research [9],
and the incidence insurance claims rate (2009–2012) for
mental health problems were not significantly different
between participating and non-participating companies.
The workplaces were very diverse in terms of their pro-
ducts, services, and markets, with 19 in manufacturing and
44 in the service sector. Of the participating workplaces, 22
were unionised, and workplaces’ workforce ranged
25–1,900 employees (average of 247.1 workers/work-
place). In each workplace, researchers first sent a com-
munication to inform all employees about the research
project. Then, a random sample of employees was selected
and invited by the researchers to individually complete a
questionnaire on company time (excluding lunch and break
times) using a touch-screen monitor that helped reducing
questionnaire’ completion time. Questionnaire administra-
tion was supervised and supported by onsite trained
research assistants. Participating workers signed an
informed consent beforehand and were given the necessary
instructions. Overall, 2,162 employees agreed to participate
in the survey (response rate 71.3 %, range 51.2–100 %)
and were employed as managers (9.7 %), supervisors
(6.8 %), professionals (15.3 %), semi-professionals/tech-
nicians (15.4 %), office workers (27.2 %), skilled labourers
(5.4 %) and unskilled/manual workers (20.2 %). After
deleting cases with missing values, the available worker
sample size was n = 1,954 employed individuals. The
study protocol was approved by the ethical committees of
the University of Montreal, McGill University, Laval
University, Bishop’s University, and Concordia University.
Measures
Mental health
Psychological distress was measured with the General
Health Questionnaire (GHQ) short-form, 12-item scale
[54] (a = 0.85), and depression with the Beck Depression
Inventory (BDI) 21-item scale [11] (a = 0.91). Emotional
exhaustion was assessed with five items from the Maslach
Burnout Inventory (MBI) general survey [64] (a = 0.90).
Emotional exhaustion was retained because it is widely
viewed as the most representative of the burnout syndrome
[44, 53, 65], and a recent study showed emotional
exhaustion, compared to cynicism and professional effi-
cacy, to be the most important component correlated with
workers diurnal cortisol profiles [51].
Workplace
Skill utilisation, decision authority, psychological
demands, and social support from colleagues and the
supervisor were derived from the Job Content Question-
naire [38]. Responses were based on a 4-point Likert scale
(strongly disagree-strongly agree). Skill utilisation con-
sisted of six items (ex: my job requires that I learn new
things, a = 0.80). Decision authority contained three items
(ex: my job allows me to make a lot of decisions on my
own, a = 0.79). Psychological demands were measured by
nine items (ex: my job requires working very fast,
a = 0.73). Social support from colleagues was measures
with four items (ex: the people I work with are helpful in
getting the job done, a = 0.83), and four items for the
support from the supervisor (ex: my supervisor is helpful in
getting the job done, a = 0.89). Physical demands, rec-
ognition, career perspectives and job insecurity were
derived from the Effort–Reward Imbalance questionnaire
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
[67]. Responses were based on a 4-point Likert scale
(strongly disagree-strongly agree). Physical demands were
based on a single item (my job is physically demanding).
Recognition contained six items (ex: I receive the respect I
deserve from my superiors, a = 0.82), career perspectives
4 items (ex: my job promotion prospects are poor, reverse
coding, a = 0.69), and job insecurity 2 items (ex: I have
experienced or I expect to experience an undesirable
change in my work situation, a = 0.65). Abusive super-
vision was measured with 15 items from Tepper abusive
supervision questionnaire [76] (ex: tells me my thoughts or
feelings are stupid, a = 0.91). Responses are based on a
5-point scale (1 = I cannot remember him ever using this
behaviour with me, 6 = He uses this behaviour very often
with me). Interpersonal conflicts during the previous
12 months contained five items from Harvey and col-
leagues questionnaire (ex: have you had an argument with
someone, a = 0.80). Response are based on a 4 point scale
(1 = never, 4 = very often) [36]. Workplace harassment
was measured using three 4-point Likert indicators
(1 = never, 4 = very often) from the Quebec Health and
Social Survey [25]. The respondent was to indicate whe-
ther, during the previous 12 months, he or she had been
subjected to physical violence or intimidation and/or been
the object of unwelcome remarks or actions of a sexual
nature in the workplace.
Family
Marital status was coded 1 for people married or living in a
civil union and 0 for others. Parental status measured the
presence (yes/no) of minor children in the household.
Household income was determined using a 10-point ordinal
scale (1 = less than $20,000, 12 = $120,000 and more).
Marital strains was assessed with four binary items (false-
true, no-yes) taken from Wheaton (ex: your partner doesn’t
understand you, a = 0.70) [82]. Parental strains had three
items (false/true) taken from Wheaton (ex: a child’s
behaviour is a source of serious concern to you, a = 0.60)
[82]. Work–Family conflict was measured with the Gutek
and colleagues instrument with responses based on a
5-point scale (strongly disagree-strongly agree) that dis-
tinguish both directions of the conflict [32]; the first one
being work-to-family spillover (four items, ex: my work
takes up time that I’d like to spend with family/friends,
a = 0.79) and the second family-to-work spillover (4
items, ex: I’m often too tired at work because of the things
I have to do at home, a = 0.74).
Social network
Social support outside the workplace was based on four
items (no–yes) from the Statistics Canada National
Population Health Survey [20] asking respondents if they
had a confidant, someone to count on in a crisis situation,
someone to count on when making personal decisions,
someone who makes them feel loved and cared for. The
scale was dichotomised as low (0 = 0–3) and high
Table 1 Sample descriptive statistics
Mean/
proportion
SD Min–
max
Mental health
Psychological distress 2.16 2.62 0–12
Depression 7.10 7.13 0–54
Emotional exhaustion 1.69 1.36 0–6
Work
Skill utilisation 17.72 3.38 6–24
Decision authority 8.62 2.00 3–12
Physical demands 2.01 0.97 1–4
Psychological demands 23.44 3.86 10–36
Working hours 40.64 10.75 6.5–168
Irregular work schedule 1.51 0.79 1–4
Support colleagues 12.51 1.95 4–16
Support supervisor 11.91 2.59 4–16
Abusive supervision 18.56 6.34 15–69
Interpersonal conflicts 7.38 2.23 2–20
Harassment 3.24 0.63 1–9
Recognition 15.67 2.63 5–20
Career perspective 10.36 2.39 4–16
Job insecurity 3.79 1.31 2–8
Family
Marital status (in couple) 0.69 0–1
Presence of minor children 0.48 0–1
Household income 6.93 3.38 1–12
Marital strains 0.44 0.90 0–4
Parental strains 0.21 0.57 0–3
Family–work conflicts 8.19 2.82 4–20
Work–family conflicts 9.90 3.50 4–20
Network
Social support (outside
work)
0.82 0–1
Agents
Gender (female) 0.49 0–1
Age 40.81 10.92 17
Physical health 1.06 1.30 0–10
Alcohol 5.58 7.73 0–80
Smoking 2.85 6.48 0–60
Physical activities 4.13 2.06 1–7
Self-esteem 19.36 3.45 2–24
Internal locus of control 19.51 4.58 0–28
Childhood stressful events 1.16 1.31 0–7
n = 1,954
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
(1 = 4) social support in order to correct for high
asymmetry.
Agent characteristics
Gender was coded 0 for males and 1 for females. Age was
measured in years starting at Cycle 1 and was indexed two
years for each subsequent cycle. Physical health was a
count of the number of physical health problems from a list
of 29 possibilities (ex: heart problems, cancer, arthritis,
etc.). Self-esteem (a = 0.87) was measured with Rosen-
berg’s 5-point (disagree–agree), six-item scale [63], and
internal locus of control (a = 0.84) with Pearlin and
Schooler’s 5-point (disagree–agree), seven-item scale [58].
Alcohol intake was measured using the summation of daily
drinks consumed over the last week (Canadian standard
drink equivalents for beer, wine, and spirits). Smoking was
based on a count of the weekly number of cigarettes and
physical activity was a measure of the monthly frequency
of one or more physical activities over 15 min in duration.
Stressful childhood life events were determined by a count
of Wheaton’s 2-point (no–yes) 7 items of events happening
before age 18 (ex: 2 weeks at the hospital, parental divorce,
parents’ alcohol or drug abuse, etc.) [82].
Table 1 presents descriptive statistics of the study
sample.
Analysis
Multilevel regression analyses [30, 70] were conducted on
data following a hierarchical structure in which workers
(n1 = 1,954) were nested within workplaces (n2 = 63).
The first multilevel regression model determined the
overall mean of psychological distress, depression, and
emotional exhaustion, as well as their variability by indi-
vidual and workplace. Next, the variables concerning the
work situation were introduced into the equation in order to
verify their contributions to the three outcomes before
controlling for family situation, social network and per-
sonality of the agent. The variables describing family,
social network, and the personality of the agent were then
entered by group, then together, in order to determine
whether the effects of the workplace were modified by the
other structures for daily life and/or the personality of the
agent. The model parameters were estimated by the itera-
tive generalised least-squares method (IGLS) using
MLwiN 2.26 [61]. In the analyses, all independent vari-
ables were centred on their respective means, with the
exception of the dichotomous variables. Finally, because of
the number of variables embedded in the regression ana-
lysis, all p-values were corrected for multiple testing with
the Benjamin and Hochberg method [14].
Results
Table 2 presents bivariate correlations between the study’s
variables. All mental health outcomes are correlated, with a
stronger correlation between psychological and depression
followed by depression and emotional exhaustion.
Psychological distress
In Table 3, Model 1 reports the overall mean of psycho-
logical distress, and the results reveal significant variation
of psychological distress at both workers and workplaces
levels. The intraclass correlation (q) indicates that work-
places accounted for 1.2 % of the total variation in psy-
chological distress.
Model 2 reports the associations for work variables and
indicates statistical significance for skill utilisation, deci-
sion authority, psychological demands, support from
supervisor, abusive supervision, work recognition, and job
insecurity. Model 3 controls for the contribution of family
situation factors, and psychological demands and support
from the supervisor are no longer significant. Model 4
controls for social support outside the workplace that
suppresses interpersonal conflicts. In Model 5, agent per-
sonality is controlled for and the results reveal that skill
utilisation, decision authority, and job recognition are no
longer significant. Model 6 contains all variables and
shows that abusive supervision and job insecurity are
associated with psychological distress. Moreover, some
family-related variables (living in couple, marital and
parental stress, work–family conflicts), and the personality
of the agent (sex, physical health status, alcohol con-
sumption, physical activity, self-esteem, internal locus of
control) were also statistically significant. This last model
explained 37 % of the variation in psychological distress
between workplaces and 32 % between workers. From the
1.2 % of variance that was between workplaces, the per-
centage of variation was now 0.9 % (p \ 0.05).
Depression
Using the same modelling approach, the results of Model 1
(Table 4) show significant variation in depression at both
the worker and workplace levels. Workplaces accounted
for 1.0 % of the total variation in depression.
Model 2 indicates statistical significance for skill util-
isation, psychological demands, support from colleagues
and supervisor, abusive supervision, interpersonal conflicts,
work recognition, career perspective, and job insecurity.
Model 3 includes family situation and psychological
demands and support from colleague and the supervisor,
job recognition and career perspective are no longer sig-
nificant. Family situation also suppresses physical
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
Ta
ble
2C
orr
elat
ion
mat
rix
of
the
stu
dy
’sv
aria
ble
s(P
ears
on
’sr)
Psy
cho
log
ical
dis
tres
s1
.00
Dep
ress
ion
0.7
41
.00
Ex
hau
stio
n0
.49
0.6
01
.00
Sk
ill
uti
lisa
tio
n-
0.2
0-
0.2
6-
0.2
21
.00
Dec
isio
nau
tho
rity
-0
.21
-0
.24
-0
.23
0.6
31
.00
Ph
ysi
cal
dem
and
s0
.02
0.0
50
.08
-0
.10
-0
.12
1.0
0
Psy
cho
log
ical
dem
and
s0
.17
0.2
00
.39
0.2
00
.06
-0
.02
1.0
0
Wo
rkin
gh
ou
rs0
.04
0.0
30
.06
0.1
10
.10
0.0
40
.15
1.0
0
Irre
gu
lar
wo
rksc
hed
ule
0.0
50
.07
0.1
50
.13
0.0
80
.11
0.2
20
.13
1.0
0
Su
pp
ort
coll
eag
ues
-0
.18
-0
.24
-0
.26
0.2
70
.24
-0.1
1-
0.1
30
.00
-0
.05
1.0
0
Su
pp
ort
sup
erv
iso
r-
0.1
8-
0.2
5-
0.3
10
.31
0.3
6-
0.1
5-
0.1
70
.00
-0
.04
0.3
71
.00
Ab
usi
ve
sup
erv
isio
n0
.24
0.3
10
.35
-0
.16
-0
.24
0.1
60
.22
0.0
50
.05
-0
.21
-0
.52
1.0
0
Inte
rper
son
alco
nfl
icts
0.2
40
.29
0.3
5-
0.1
2-
0.1
20
.14
0.3
00
.11
0.1
5-
0.2
7-
0.2
80
.44
1.0
0
Har
assm
ent
0.1
80
.20
0.2
6-
0.0
9-
0.1
40
.18
0.1
80
.08
0.1
4-
0.1
8-
0.1
90
.35
0.4
81
.00
Rec
og
nit
ion
-0
.29
-0
.35
-0
.40
0.3
20
.35
-0
.16
-0
.24
-0
.01
-0
.07
0.5
10
.62
-0
.49
-0
.40
-0
.30
1.0
0
Car
eer
per
spec
tiv
e-
0.2
3-
0.2
9-
0.2
90
.44
0.3
8-
0.1
2-
0.1
00
.04
0.0
00
.31
0.4
0-
0.3
0-
0.2
2-
0.1
50
.51
1.0
0
Job
inse
curi
ty0
.27
0.2
90
.37
-0
.17
-0
.21
0.1
00
.21
-0
.01
0.0
9-
0.2
7-
0.3
40
.28
0.2
60
.23
-0
.47
-0
.32
1.0
0
Mar
ital
stat
us
(in
cou
ple
)-
0.1
0-
0.0
9-
0.0
40
.12
0.0
9-
0.0
60
.03
0.0
2-
0.0
10
.04
0.0
2-
0.0
7-
0.0
8-
0.1
00
.06
0.0
70
.00
Pre
sen
ceo
fm
ino
rch
ild
ren
-0
.01
-0
.03
-0
.06
0.0
80
.06
-0
.04
0.0
5-
0.0
10
.00
0.0
5-
0.0
1-
0.0
1-
0.0
4-
0.0
4-
0.0
10
.02
0.0
2
Ho
use
ho
ldin
com
e-
0.0
9-
0.1
4-
0.0
30
.28
0.3
0-
0.2
60
.21
0.1
10
.12
0.0
60
.06
-0
.11
-0
.05
-0
.02
0.0
60
.16
0.0
4
Mar
ital
stra
ins
0.2
10
.23
0.1
4-
0.0
3-
0.0
2-
0.0
20
.08
-0
.02
0.0
7-
0.0
8-
0.0
50
.07
0.0
70
.01
-0
.10
-0
.09
0.1
0
Par
enta
lst
rain
s0
.14
0.1
40
.08
0.0
00
.00
0.0
20
.07
0.0
1-
0.0
1-
0.0
2-
0.0
30
.02
0.0
60
.08
-0
.07
-0
.06
0.0
9
Fam
ily
–w
ork
con
flic
ts0
.28
0.3
20
.22
-0
.13
-0
.12
0.0
50
.09
-0
.04
-0
.01
-0
.13
-0
.06
0.1
40
.16
0.0
5-
0.1
9-
0.1
50
.17
Wo
rk–
fam
ily
con
flic
ts0
.34
0.4
00
.56
0.0
0-
0.0
50
.04
0.4
50
.14
0.2
4-
0.2
0-
0.2
00
.21
0.2
50
.18
-0
.30
-0
.16
0.3
2
So
cial
sup
po
rt-
0.1
7-
0.2
4-
0.1
20
.08
0.0
5-
0.0
2-
0.0
4-
0.0
4-
0.0
50
.13
0.1
7-
0.1
1-
0.0
5-
0.0
40
.15
0.1
1-
0.1
0
Gen
der
(fem
ale)
0.0
90
.09
0.0
9-
0.1
0-
0.1
0-
0.3
00
.05
-0
.13
-0
.08
0.0
10
.02
-0
.05
0.0
10
.07
0.0
4-
0.0
10
.00
Ag
e-
0.0
7-
0.1
0-
0.0
80
.15
0.1
0-
0.0
80
.04
0.0
40
.02
-0
.04
-0
.05
-0
.02
-0
.10
-0
.02
0.0
00
.09
0.0
1
Ph
ysi
cal
hea
lth
0.1
90
.27
0.2
5-
0.0
9-
0.0
9-
0.0
20
.13
-0
.02
0.0
3-
0.1
3-
0.0
80
.13
0.2
10
.15
-0
.15
-0
.10
0.1
0
Alc
oh
ol
0.0
50
.09
0.0
60
.03
0.0
00
.10
0.0
20
.05
0.0
3-
0.0
1-
0.1
00
.07
0.0
8-
0.0
3-
0.0
7-
0.0
30
.04
Sm
ok
ing
0.1
00
.16
0.1
1-
0.0
7-
0.1
10
.18
0.0
00
.03
0.0
4-
0.0
8-
0.0
80
.09
0.0
60
.01
-0
.08
-0
.09
0.0
6
Ph
ysi
cal
acti
vit
ies
-0
.14
-0
.16
-0
.07
0.1
20
.10
-0
.10
0.0
90
.01
0.0
40
.01
0.0
6-
0.0
30
.00
-0
.01
0.0
40
.04
-0
.02
Sel
f-es
teem
-0
.34
-0
.45
-0
.27
0.3
00
.26
-0
.04
-0
.03
-0
.02
0.0
00
.24
0.1
8-
0.0
9-
0.1
1-
0.0
40
.28
0.2
2-
0.2
0
Inte
rnal
locu
so
fco
ntr
ol
-0
.45
-0
.56
-0
.40
0.3
20
.30
-0
.12
-0
.10
0.0
2-
0.0
20
.29
0.2
7-
0.2
5-
0.2
3-
0.1
60
.40
0.2
8-
0.3
0
Ch
ild
ho
od
stre
ssfu
lev
ents
0.1
70
.23
0.1
7-
0.1
2-
0.1
20
.09
0.0
70
.02
0.0
2-
0.0
9-
0.0
70
.14
0.1
80
.13
-0
.13
-0
.16
0.0
9
Mar
ital
stat
us
(in
cou
ple
)1
.00
Pre
sen
ceo
fm
ino
rch
ild
ren
0.2
81
.00
Ho
use
ho
ldin
com
e0
.43
0.2
31
.00
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
demands. Model 4 controls for social support outside the
workplace. Support for colleagues is now not significant,
and social support outside the workplace suppresses deci-
sion authority. In Model 5, personality of the agent is
controlled for and the results reveal that skill utilisation,
support from colleague and the supervisor, interpersonal
conflicts and job recognition are now not statistically sig-
nificant, and agent personality suppresses physical
demands. Model 6 contains all variables and shows that
only physical and psychological demands and abusive
supervision are associated with depression. Some family-
related variables (marital status, marital and parental
strains, work-family conflicts), and the personality of the
agent (all variables) were also statistically significant. This
last model explained 61 % of the variation in depression
between workplaces and 48 % between workers. Overall,
variation of depression between workplaces is no longer
significant.
Emotional exhaustion
Table 5 presents the results for emotional exhaustion.
Model 1 shows significant variation of the outcome at
both the worker and workplace levels. Workplaces
accounted for 5.2 % of the total variation in emotional
exhaustion. Model 2 reports associations for work vari-
ables and indicated statistical significance for skill util-
isation, psychological demands, irregular work schedule,
abusive supervision, interpersonal conflicts, harassment,
and job insecurity. Model 3 controls for family situation
and irregular work schedule and harassment lost their
significance. Model 4 controls for social support outside
the workplace, and harassment is now no longer signifi-
cant. In Model 5, personality of the agent is taken into
account and the results reveal no specific modifications
regarding the association between work variables an
emotional exhaustion. Model 6 contains all variables and
shows that skill utilisation, psychological demands, abu-
sive supervision, interpersonal conflicts and job insecurity
are associated with emotional exhaustion. Some family-
related variables (parental status, family–work conflicts,
work-family conflicts), and the personality of the agent
(age, physical health, internal locus of control) were also
statistically significant. Model 6 explained 69 % of the
variation in emotional exhaustion between workplaces and
49 % between workers. The percentage of variation left for
workplaces was 1.8 % (p \ 0.01).
At the end, multicollinearity tests were conducted
because of correlated independent variables in multilevel
regression models. Based on Model 6 of each outcome, the
variance inflation factor (VIF) ranged 1.10–2.58 with an
average of 1.49. These values are largely below theTa
ble
2co
nti
nu
ed
Mar
ital
stra
ins
0.3
30
.16
0.0
91
.00
Par
enta
lst
rain
s0
.09
0.3
80
.09
0.1
61
.00
Fam
ily
–w
ork
con
flic
ts-
0.0
70
.07
-0
.13
0.1
90
.17
1.0
0
Wo
rk–
fam
ily
con
flic
ts0
.02
0.0
50
.12
0.1
60
.12
0.3
81
.00
So
cial
sup
po
rt-
0.0
1-
0.0
40
.02
-0
.17
-0
.09
-0
.11
-0
.15
1.0
0
Gen
der
(fem
ale)
-0
.06
0.0
10
.04
-0
.01
0.0
5-
0.0
10
.07
0.0
81
.00
Ag
e0
.10
0.1
00
.21
-0
.01
0.0
9-
0.1
60
.01
-0
.07
0.0
81
.00
Ph
ysi
cal
hea
lth
0.0
0-
0.0
2-
0.0
20
.07
0.0
40
.08
0.1
7-
0.0
60
.18
0.1
01
.00
Alc
oh
ol
-0
.02
-0
.07
-0
.01
0.0
0-
0.0
30
.04
0.0
10
.00
-0
.24
-0
.05
-0
.03
1.0
0
Sm
ok
ing
-0
.07
-0
.02
-0
.21
0.0
60
.02
0.1
10
.07
-0
.06
-0
.11
-0
.03
0.0
30
.22
1.0
0
Ph
ysi
cal
acti
vit
ies
0.0
1-
0.0
40
.17
-0
.07
-0
.02
-0
.07
-0
.05
0.0
80
.05
-0
.02
-0
.07
0.0
5-
0.1
91
.00
Sel
f-es
teem
0.0
7-
0.0
20
.11
-0
.14
-0
.13
-0
.27
-0
.22
0.2
1-
0.0
50
.05
-0
.13
-0
.04
-0
.06
0.1
11
.00
Inte
rnal
locu
so
fco
ntr
ol
0.0
70
.02
0.2
0-
0.2
1-
0.1
4-
0.3
7-
0.3
50
.23
-0
.02
0.0
2-
0.1
8-
0.0
4-
0.1
20
.19
0.5
41
.00
Ch
ild
ho
od
stre
ssfu
lev
ents
-0
.07
-0
.06
-0
.15
0.0
70
.05
0.0
90
.11
-0
.05
0.0
6-
0.1
10
.23
0.0
20
.12
-0
.03
-0
.09
-0
.17
1.0
0
r\
-0
.06
and
r[
0.0
6ar
ep\
0.0
1
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
Table 3 Results of multilevel regression modelling of psychological distress (unstandardised coefficients)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 2.168** 2.157** 2.672** 2.807** 2.003** 2.671**
Work
Skill utilisation -0.075** -0.061* -0.072** -0.016 -0.018
Decision authority -0.111** -0.095* -0.118** -0.038 -0.057
Physical demands -0.117 -0.164* -0.111 -0.129* -0.135
Psychological demands 0.066** 0.022 0.067* 0.050** 0.026
Working hours 0.008 0.008 0.007 0.010 0.009
Irregular work schedule 0.063 -0.030 0.050 0.075 -0.003
Support colleagues -0.038 -0.005 -0.028 0.019 0.023
Support supervisor 0.085** 0.044 0.094** 0.064* 0.052
Abusive supervision 0.035** 0.027* 0.033** 0.030** 0.028*
Interpersonal conflicts 0.083 0.053 0.089** 0.040 0.029
Harassment 0.151 0.147 0.147 0.174 0.160
Recognition -0.095** -0.061 -0.091** -0.035 -0.024
Career perspective -0.046 -0.021 -0.043 -0.037 -0.019
Job insecurity 0.259** 0.181** 0.248** 0.188** 0.153**
Family
Marital status (in couple) -0.600** -0.543**
Presence of minor children -0.177 -0.117
Household income -0.029 0.006
Marital strains 0.496** 0.358**
Parental strains 0.365** 0.259*
Family–work conflicts 0.084** 0.034
Work–family conflicts 0.135** 0.083**
Network
Social support -0.782** -0.251
Agents
Gender (female) 0.330* 0.295*
Age -0.012* -0.010
Physical health 0.131** 0.110*
Alcohol 0.016* 0.016*
Smoking 0.011 0.005
Physical activities -0.104** -0.090**
Self-esteem -0.092** -0.074**
Internal locus of control -0.144** -0.110**
Childhood stressful events 0.084 0.072
Random part
Workplaces variance 0.081** 0.043 0.025 0.045* 0.062* 0.042*
Workers variance 6.797** 5.652** 5.123** 5.569** 4.797** 4.624**
q 0.012 0.008 0.005 0.008 0.013 0.009
Goodness-of-fit
v2 670.35** 1,445.65** 700.72** 1,183.99** 1,738.08**
df 14 21 15 23 31
R2 (workplaces) 0.250 0.368 0.252 0.278 0.366
R2 (workers) 0.172 0.252 0.184 0.294 0.322
Benjamin and Hochberg method corrected p values for multiple testing
* p \ 0.05; ** p \ 0.01
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
Table 4 Results of multilevel regression modelling of depression (unstandardised coefficients)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 7.097** 7.098** 8.202** 9.673** 6.649** 8.671**
Work
Skill utilisation -0.321** -0.278** -0.309** -0.085 -0.098
Decision authority -0.176 -0.098 -0.200* 0.086 0.038
Physical demands -0.208 -0.379* -0.181 -0.317* -0.315*
Psychological demands 0.219** 0.070 0.222** 0.167** 0.089*
Working hours 0.010 0.008 0.006 0.012 0.006
Irregular work schedule 0.245 -0.035 0.194 0.262 0.044
Support colleagues -0.206* -0.097 -0.165 0.007 0.026
Support supervisor 0.173* 0.055 0.210** 0.061 0.058
Abusive supervision 0.145** 0.118** 0.139** 0.126** 0.123**
Interpersonal conflicts 0.280** 0.200* 0.303** 0.100 0.081
Harassment 0.280 0.237 0.254 0.332 0.249
Recognition -0.221* -0.138 -0.205* -0.006 0.007
Career perspective -0.181* -0.109 -0.166* -0.137* -0.103
Job insecurity 0.616** 0.314** 0.572** 0.321** 0.188
Family
Marital status (in couple) -1.031** -0.752*
Presence of minor children -0.799* -0.534
Household income -0.179** -0.029
Marital strains 1.271** 0.701**
Parental strains 1.030** 0.576*
Family–work conflicts 0.240** 0.048
Work–family conflicts 0.497** 0.284**
Network
Social support -3.096** -1.492**
Agents
Gender (female) 0.935** 0.892**
Age -0.050** -0.050**
Physical health 0.618** 0.577**
Alcohol 0.057** 0.060**
Smoking 0.072** 0.057**
Physical activities -0.257** -0.222**
Self-esteem -0.421** -0.360**
Internal locus of control -0.480** -0.382**
Childhood stressful events 0.387** 0.334**
Random part
Workplaces variance 0.525 0.243* 0.013 0.199 0.090 0.000
Workers variance 50.322** 38.927** 33.696** 37.683** 27.503** 26.236**
q 0.010 0.006 0.000 0.005 0.003 0.000
Goodness-of-fit
v2 897.32 2,005.49 967.11 1,871.98 2,634.86
df 14 21 15 23 31
R2 (workplaces) 0.302 0.488 0.342 0.545 0.606
R2 (workers) 0.230 0.337 0.255 0.457 0.484
Benjamin and Hochberg method corrected p values for multiple testing
* p \ 0.05; ** p \ 0.01
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
Table 5 Results of multilevel regression modelling of emotional exhaustion (unstandardised coefficients)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 1.674** 1.677** 1.742** 1.796** 1.593** 1.638**
Work
Skill utilisation -0.061** -0.059** -0.060** -0.039** -0.046**
Decision authority -0.026 -0.016 -0.027 0.001 -0.003
Physical demands -0.006 -0.026 -0.005 -0.005 -0.018
Psychological demands 0.102** 0.059** 0.102** 0.094** 0.058**
Working hours 0.002 -0.002 0.002 0.003 -0.001
Irregular work schedule 0.095** 0.021 0.092* 0.099** 0.033
Support colleagues -0.025 -0.009 -0.023 -0.007 -0.004
Support supervisor 0.000 -0.006 0.002 -0.009 -0.007
Abusive supervision 0.021** 0.021** 0.021** 0.018** 0.021**
Interpersonal conflicts 0.051** 0.045** 0.052** 0.034* 0.031*
Harassment 0.133* 0.084 0.132 0.129* 0.083
Recognition -0.028 -0.013 -0.027 -0.011 -0.002
Career perspective -0.025 -0.020 -0.025 -0.019 -0.019
Job insecurity 0.146** 0.088** 0.144** 0.126** 0.086**
Family
Marital status (in couple) 0.046 0.063
Presence of minor children -0.210** -0.194**
Household income -0.020 -0.008
Marital strains 0.062 0.029
Parental strains 0.076 0.062
Family–work conflicts -0.016 -0.032**
Work–family conflicts 0.153** 0.138**
Network
Social support -0.144 0.034
Agents
Gender (female) 0.176** 0.115
Age -0.008** -0.008**
Physical health 0.106** 0.094**
Alcohol 0.006 0.007
Smoking 0.008 0.005
Physical activities -0.025 -0.018
Self-esteem -0.021* -0.017
Internal locus of control -0.050** -0.029**
Childhood stressful events 0.012 0.006
Random part
Workplaces variance 0.095** 0.022** 0.016** 0.022** 0.022** 0.017**
Workers variance 1.750** 1.204** 0.982** 1.201** 1.084** 0.927**
q 0.052 0.018 0.016 0.018 0.020 0.018
Goodness-of-fit
v2 973.52 1,754.94 978.28 1,308.69 1,927.96
df 14 21 15 23 31
R2 (workplaces) 0.595 0.683 0.597 0.623 0.687
R2 (workers) 0.335 0.459 0.337 0.400 0.488
Benjamin and Hochberg method corrected p values for multiple testing
* p \ 0.05; ** p \ 0.01
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
threshold of 4 (sign) or 10 (serious) indicating multicol-
linearity problems [13].
Discussion
This study has examined the contribution of the multilevel
determinants of workers’ mental health model, with an
emphasis on the role of the workplace context and work
organisation conditions on psychological distress, depres-
sion, and emotional exhaustion, when non-work and indi-
vidual factors are taken into account. The results support
the relevance of the proposed model that views workers’
mental health as being the product of stress caused by
constraints-resources brought to bear simultaneously by
structures of daily life and agent personality. As work is
concerned, the workplace level per se, accounts for a small
part of the variation in psychological distress (1.2 %),
depression (1.0 %) and emotional exhaustion (5.2 %).
Moreover, when indicators of constraints-resources in
structures of daily life and the personality of the agent are
controlled for, the variation in depression across work-
places is no longer significant, and only 0.9 % of the var-
iation in psychological distress, and 1.8 % in emotional
exhaustion remained between workplaces. Therefore,
workplaces are not strongly differentiating themselves on
the level of mental health symptoms reported by their
employees. Even if these small variances at the workplace
level may be attributable to the relatively small numbers of
workplaces (n = 63), the present study suggests that psy-
chological distress, and depression symptoms levels are
mostly comparable across organisations analysed here.
Nevertheless, emotional exhaustion symptoms, as the
major component of the burnout process, seems to be more
elevated in specific firms, since larger variations in emo-
tional exhaustion symptoms between workplaces were
observed compared to psychological distress and
depression.
Workers evaluation of work organisation conditions
prove to be important within workplace constraints-
resources associated with the three mental health outcomes
studied here. Constraints-resources were, however, asso-
ciated differently with the three outcomes. As for task
design, a higher level of skill utilisation is associated with
lower levels of depression and emotional exhaustion. This
is consistent with previous studies [37, 62, 72] and high-
lights the importance of designing task that motivate and
challenge the skill of workers. However, the level of
decision authority is not associated with outcomes when
other structures of daily life and personality of the agent are
taken into account. As for work demands, work hours and
schedule are not significant when controlled for family
situation, social support outside the workplace and
individual characteristics. Higher physical demands are
surprisingly related to a lower level of depression symp-
toms. Such a result may be explained by the way physical
demands were measured, as they were indexed with only
one item. It did not measure exposure to physical risk
(exposure to contaminants, dust, lifting heavy objects, etc.),
but more of the physical energy workers invested in their
task. Such investment may require a better physical and
general health to perform the job. For psychological
demands, they are associated with higher level of depres-
sion and emotional exhaustion symptoms. This is consis-
tent with previous studies [37, 59], but not with
psychological distress [72] when other structures of daily
life are accounted for. Regarding social relations, abusive
supervision clearly appears as an important stressor asso-
ciated with higher levels of psychological distress,
depression, and emotional exhaustion. Such a style of
supervision, based on sustained display of hostile verbal
and nonverbal behaviours, excluding physical contact [77],
is detrimental for mental health because supervisors are in
relationship with subordinates on a day-to-day basis, and
thus constituted a chronic stressor for those workers
exposed to it. We also found that interpersonal conflicts at
work were associated with a high level of emotional
exhaustion symptoms. Social support from colleagues and
the supervisor, as well as harassment, were not significant
after controlling for non-work factors and personal char-
acteristics. Concerning work gratifications, our results
confirmed, as some previous studies did [15, 18, 74], that
job insecurity is also a stressful condition associated with
elevated levels of psychological distress and emotional
exhaustion. However, neither job recognition nor career
perspectives associate with the three outcomes when fam-
ily situation, social support outside the workplace, and
individual characteristics are controlled for.
Results obtained here support the theoretical model
concerning the stressful role of constraints-resources
embedded in the other structures of daily life that include
family and social network outside the workplace. These
structures promote life experiences and the possibilities for
self-realisation, but they could also be sources of stress. All
constraints-resources with the family associate with any
one of the three outcomes, to the exception of family
income. Less mental health symptoms were found for those
living in couple (psychological distress, depression), hav-
ing children in the household (emotional exhaustion), and
experiencing family-to-work conflicts (emotional exhaus-
tion). However, stressful marital and parental relationships
within the household are marked by higher levels of psy-
chological distress and depression. As for work-to-family
conflicts, they appear to be an important chronic stressor,
because they relate to more symptoms for all mental health
outcomes analysed here. Family situation also modified the
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
relationship between work stressors and the three out-
comes. When family situation is controlled for in the
analysis, psychological distress is no longer associated with
psychological demands and support from the supervisor,
depression is no longer associated with psychological
demands and social support from both colleagues and
supervisor, and emotional exhaustion is no longer associ-
ated with irregular work schedule and harassment. Results
in Table 2 may suggest the apparent confounding effect to
be attributable to work-family conflicts as they are posi-
tively correlated with psychological demands, irregular
work schedule and harassment, while negatively associated
with support from colleagues and supervisor.
Concerning social support outside the workplace, it is
associated with less symptoms of depression, while the
relationships with psychological distress and emotional
exhaustion were not significant. Furthermore, when it is
controlled for, support from colleague is no longer signif-
icant with depression, and harassment lost its significant
effect on emotional exhaustion. Social support outside the
workplace seems to confound these associations, because it
is associated with higher level of support from the super-
visor, while there is as a small tendency to be related to less
harassment (see Table 2).
The results also confirm the role of the agent personality
in the theoretical model. Workers are not passive beings
who are subservient to the social conditions in which they
live since when they act, they bring with them constraints
and resources peculiar to them and shaped by their bodies,
minds, and social context. Being a woman was related to
more symptoms of psychological distress and depression,
while age was related to lower symptoms of emotional
exhaustion. Physical health problem on its side associated
with more symptoms for all outcomes, alcohol intake with
higher levels of psychological distress and depression, and
smoking with more symptoms of depression. Fewer
symptoms of psychological distress and depression were
found with higher level of physical activities, while self-
esteem (psychological distress, depression) and internal
locus of control (emotional exhaustion) were negatively
associated with the outcomes. Also, agent personality
apparently contributes to the confounding of skill utilisa-
tion (psychological distress, depression), decision authority
(psychological distress), support from colleagues and
supervisor (depression), and job recognition (psychological
distress, depression). According to correlations of Table 2,
the possible confounding effect seems to be attributable to
personality traits, because self-esteem and internal locus of
control are associated with higher levels of skill utilisation,
decision authority, support from colleagues and supervi-
sion as well with a higher level of job recognition.
The present study nevertheless has limitations. Firstly,
the data are cross-sectional, which implies that the observed
relationships cannot be interpreted causally and will need to
be replicated longitudinally. Some reverse causation might
be possible, as workers suffering from mental symptoms
may negatively evaluate their work conditions. Secondly,
we cannot rule out the possibility of a common method
variance bias, because all measurements were based on one
source. However, workers came from 63 different firms,
reducing therefore the bias attributable to measurements
based on one specific context [60]. We also conducted a
single unmeasured latent factor analysis with Mplus 7.11 as
suggested by Podsakoff et al. [60] to verify if the latent
factor accounted for variance and covariance between
measurements. Results gave v2 = 19,193.70, df = 528,
p \ 0.0001, suggesting the common method variance bias
to be small. Thirdly, results cannot be generalised to the
overall workforce as data came from a single insurance
company, but the 63 firms sampled were diversified in
terms of economic sectors, firm sizes, and unionisation.
Fourthly, the companies’ response rate of 41 % may also
have introduced a selection bias, such as company experi-
encing more problems with workers’ mental health might
have been more willing to participate in the study. How-
ever, the response rate was significantly higher compared to
the ones usually found in organisational research [9], and
the incidence insurance claims rate (2009–2012) for mental
health problems were not significantly different between
respondent and non-respondent companies. Fifthly, the
analysis does not take into account workplace factors
related to the physical environment (dust, noise, cold, heat,
toxic, etc.), human resources practices, health and safety
resources or other elements in the work contract that allow
employees to better balance work and family responsibili-
ties. Sixthly, while we controlled for gender differences in
the analysis, patterns of associations may be different
between genders and would need to be investigated in
future studies. Finally, psychological distress, depression,
and emotional exhaustion are all correlated. Independent
variables cannot be tested for difference in coefficients
across the three outcome variables [3], which would have
required the use of multivariate multilevel regression
models with a larger sample size at the companies’ level.
Therefore, the contribution of independent variables for
specifically psychological distress, depression and emo-
tional exhaustion will require further studies.
Despite these limitations, this study demonstrates that
the multilevel determinants of workers’ mental health
model explain a substantial part of the variance in psy-
chological distress, depression, and emotional exhaustion.
Based on this model, pathogenic work organisation con-
ditions, as estimated here, appear more important for
emotional exhaustion symptoms compared to psychologi-
cal distress and depression. Furthermore, the results of this
study clearly demonstrate that family situation, social
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
support outside the workplace, and personal characteristics
are also in and of themselves important factors associated
with workers’ mental health. Not only they associate with
mental health, they also modulate the number and the type
of work stressors that related to mental health symptoms.
In the end, we need to consider broadening approaches
in occupational mental health to avoid coming to erroneous
conclusions about the relationship between work and
mental health. Theoretical and empirical studies must
recognise the complexity of workers’ mental health deter-
minants if we want to be better able to capture and inter-
vene on what is going wrong with work organisation
conditions experienced by the worker. This study therefore
replies to previous claims that non-work and individual
factors must be integrated in occupational mental health
research to arrive at a better understanding of workers’
mental health problems [10, 16]. For example, if non-work
or individual factors appear as the primary explanation of
mental health symptoms of workers in a particular com-
pany, interventions on work conditions that will help better
coping with stressful life conditions will have a better
chance to be successful in reducing mental health symp-
toms. More research is thus needed to help the develop-
ment of diagnosis and corrective measures that will not just
focus on the work factor, but also on other structures of
daily life people are involved in, as well as individual
characteristics on which interventions are possible.
Acknowledgments This study was supported by the Canadian
Health Research Institutes under Grant number: 200607MHF-
164381-MHF-CFCA-155960, and the Fonds de recherche du Quebec-
Sante under Grant number: 13928. The authors also thank Standard
Life Canada for their help in workplace recruitment and Marie-Eve
Blanc and Julie Dextras-Gauthier for the field work.
Conflict of interest On behalf of all authors, the corresponding
author states that there is no conflict of interest.
References
1. Adam S, Gyorffy Z, Susanszky E (2008) Physician burnout in
Hungary: a potential role for work–family conflict. J Health
Psychol 13:847–856
2. Adriaenssens J, De Gucht V, Maes S (2013) Causes and conse-
quences of occupational stress in emergency nurses, a longitu-
dinal study. J Nurs Manag. doi:10.1111/jonm.12138
3. Afifi A, Clark V, May S (2004) Computer-Aided Multivariate
Analysis, 4th edn. Chapman & Hall/CRC, Boca Raton
4. Ahola K, Honkonen T, Isometsa E, Kalimo R, Nykyri E, Ko-
skinen S, Aromaa A, Lonnqvist J (2006) Burnout in the general
population results from the Finnish Health 2000 Study. Soc
Psychiatry Psychiatr Epidemiol 41:11–17
5. Ahola K, Pulkki-Raback L, Kouvonen A, Rossi H, Aromaa A,
Lonnqvist J (2012) Burnout and behavior-related health risk
factors: results from the population-based Finnish Health 2000
study. J Occup Environ Med 54:17–22
6. Alexander JC (1987) Action and its environments. In: Alexander
JC, Giesen B, Munch R, Smelser NJ (eds) The micro–macro link.
University of California, Berkeley, pp 289–318
7. Archer MS (1995) Realist social theory. The morphogenetic
approach. Cambridge University Press, Cambridge
8. Ballard TJ, Romito P, Lauria L, Vigiliano V, Caldora M,
Mazzanti C, Verdecchia A (2006) Self perceived health and
mental health among women flight attendants. Occup Environ
Med 63:33–38
9. Baruch Y, Holtom BC (2008) Survey response rate levels and
trends in organizational research. Hum Relat 61:1139–1160
10. Beauregard N, Marchand A, Blanc M-E (2011) What do we know
about the non-work determinants of workers’ mental health? A
systematic review of longitudinal studies. BMC Public Health
11:439
11. Beck AT, Steer RA, Brown GK (1996) BDI-II. Beck depression
inventory, 2 edn. Psychological Corporation, San Antonio
12. Bellavia GM, Frone MR (2005) Work–family conflict. Handbook
of work stress, pp 113–147
13. Belsley DA, Kuh E, Welsch RE (1980) Regression diagnostics:
identifying influential data and sources of collinearity. Wiley,
New York
14. Benjamini Y, Hochberg Y (1995) Controlling the false discovery
rate: a practical and powerful approach to multiple testing. J R
Stat Soc Ser B 57
15. Blackmore ER, Stansfeld SA, Weller I, Munce S, Zagorski BM,
Stewart DE (2007) Major depressive episodes and work stress:
results from a National Population Survey. Am J Public Health
97:2088–2093
16. Bonde JPE (2008) Psychosocial factors at work and risk of
depression: a systematic review of the epidemiological evidence.
Occup Environ Med 65:438–445
17. Boyas J, Wind LH (2010) Employment-based social capital, job
stress, and employee burnout: a public child welfare employee
structural model. Child Youth Serv Rev 32:380–388
18. Burgard SA, Brand JE, House JS (2009) Perceived job insecurity
and worker health in the United States. Soc Sci Med 69:777–785
19. Campbell C (1996) The myth of social action. Cambridge Uni-
versity Press, Cambridge
20. Catlin G, Will P (1992) The National Population Health Survey:
highlights of initial developments. Health Rep 4:313–319
21. Chida Y, Steptoe A (2009) Cortisol awakening response and
psychosocial factors: a systematic review and meta-analysis. Biol
Psychol 80:265–278
22. Clark C, Pike C, McManus S, Harris J, Bebbington P, Brugha T,
Jenkins R, Meltzer H, Weich S, Stansfeld S (2012) The contri-
bution of work and non-work stressors to common mental dis-
orders in the 2007 Adult Psychiatric Morbidity Survey. Psychol
Med 42:829–842
23. Clays E, De Bacquer D, Leynen F, Kornitzer M, Kittel F, De
Backer G (2007) Job stress and depression symptoms in middle-
aged workers—prospective results from the Belstress study.
Scand J Work Environ Health 33:252–259
24. D’Auria D (1997) Stress and stress related illness. In: Brume D,
Gerhardsson G, Crockford GW, D’Auria D (eds) The workplace,
vol 1., Fundamentals of health. Safety and welfareInternational
Occupational Safety and Health Information Center; International
Labour Office, Scandinavian Science Publishers (Oslo), Geneva,
pp 954–960
25. Daveluy C, Pica L, Audet N, Courtemanche R, Lapointe F, Cote
L et al (2000) Enquete sociale et de sante 1998. Institut de la
statistique du Quebec, Quebec
26. Demerouti E, Bakker AB, Nachreiner F, Schaufeli WB (2001)
The job demands—resources model of burnout. J Appl Psychol
86:499–512
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
27. Driesen K, Jansen NW, Kant I, Mohren DC, van Amelsvoort LG
(2010) Depressed mood in the working population: associations
with work schedules and working hours. Chronobiol Int
27:1062–1079
28. Eriksen W, Tambs K, Knardahl S (2006) Work factors and
psychological distress in nurses’ aides: a prospective cohort
study. BMC Public Health 6:290
29. Giddens A (1987) La Constitution de la societe. PUF, Paris
30. Goldstein H (1995) Multilevel statistical models. Edward Arnold,
Halstead Press, London, New York
31. Grandey AA, Kern JH, Frone MR (2007) Verbal abuse from
outsiders versus insiders: comparing frequency, impact on emo-
tional exhaustion, and the role of emotional labor. J Occup Health
Psychol 12:63–79
32. Gutek AB, Searle S, Klepa L (1991) Rational versus gender role
explanations for work-family conflict. J Appl Psychol
76:560–568
33. Haines VY III, Marchand A, Rousseau V, Demers A (2008) The
mediating role of work-to-family conflict in the relationship
between shiftwork and depression. Work Stress 22:341–356
34. Haines VY 3rd, Marchand A, Harvey S (2006) Crossover of
workplace aggression experiences in dual-earner couples.
J Occup Health Psychol 11:305–314
35. Hansen AM, Hogh A, Persson R, Karlson B, Garde AH, Orbaek P
(2006) Bullying at work, health outcomes, and physiological
stress response. J Psychosom Res 60:63–72
36. Harvey S, Blouin C, Stout D (2006) Proactive personality as a
moderator of outcomes for young workers experiencing conflict
at work. Personal Individ Differ 40:1063–1074
37. Huang Y-H, Chen C-H, Du P-L, Huang I-C (2012) The causal
relationships between job characteristics, burnout, and psycho-
logical health: a two-wave panel study. Int J Human Res Manag
23:2108–2125
38. Karasek R, Gordon G, Pietrokovsky C, Frese M, Pieper C, Sch-
wartz J, Fry L, Schirer D (1985) Job content questionnaire: ques-
tionnaire and users’ guide. University of Massachusetts, Lowell
39. Karasek RA (1979) Job demands, job decision latitude, and
mental strain: implication for job redesign. Adm Sci Q
24:285–309
40. Karasek RA, Theorell T (1990) Healthy work: stress, produc-
tivity, and the reconstruction of the working life. Basic Books,
New York
41. Kleppa E, Sanne B, Tell GS (2008) Working overtime is asso-
ciated with anxiety and depression: the Hordaland Health Study.
J Occup Environ Med 50:658–666
42. Klersy C, Callegari A, Martinelli V, Vizzardi V, Navino C,
Malberti F, Tarchini R, Montagna G, Guastoni C, Bellazzi R,
Rampino T, David S, Barbieri C, Dal Canton A, Politi P,
Working Group on B, Dialysis (2007) Burnout in health care
providers of dialysis service in Northern Italy—a multicentre
study. Nephrol Dial Transplant 22:2283–2290
43. Kowalski C, Ommen O, Driller E, Ernstmann N, Wirtz MA,
Kohler T, Pfaff H (2010) Burnout in nurses—the relationship
between social capital in hospitals and emotional exhaustion.
J Clin Nurs 19:1654–1663
44. Lee RT, Ashforth BE (1996) A meta-analytic examination of the
correlates of the three dimensions of job burnout. J Appl Psychol
81:122–133
45. Lindwall M, Gerber M, Jonsdottir IH, Borjesson M, Ahlborg G
(2013) The relationships of change in physical activity with
change in depression, anxiety, and burnout: a longitudinal study
of swedish healthcare workers. Health Psychol
46. Losa Iglesias ME, de Bengoa Becerro, Vallejo R, Salvadores
Fuentes P (2010) The relationship between experiential avoid-
ance and burnout syndrome in critical care nurses: a cross-sec-
tional questionnaire survey. Int J Nurs Stud 47:30–37
47. Marchand A, Blanc M-E (2011) Occupation, work organisation
conditions and the development of chronic psychological distress.
Work 40:425–435
48. Marchand A, Demers A, Durand P (2005) Do occupation and
work conditions really matter? A longitudinal analysis of psy-
chological distress experiences among Canadian workers. Sociol
Health Illn 27:602–627
49. Marchand A, Demers A, Durand P (2005) Does work really cause
distress? The contribution of occupational structure and work
organization to the experience of psychological distress. Soc Sci
Med 60:1–14
50. Marchand A, Durand P, Lupien S (2013) Work hours and cortisol
variation from non-working to working days. Int Arch Occup
Environ Health 85:553–559
51. Marchand A, Juster R-P, Durand P, Lupien S (2014) Burnout
symptom sub-types and cortisol profiles: what’s burning most?
Psychoneuroendocrinology 40:27–36
52. Martinko MJ, Harvey P, Brees JR, Mackey J (2013) A review of
abusive supervision research. J Organ Behav 34:S120–S137
53. Maslach C (1998) A multidimensional theory of burnout. In:
Cooper C (ed) Theories of organizational stress. University Press,
Oxford, pp 68–85
54. McDowell I, Newell C (1996) Measuring health: a guide to rating
scales and questionnaires, 2nd edn. Oxford University Press, New
York
55. Merecz D, Drabek M, Moscicka A (2009) Aggression at the
workplace—psychological consequences of abusive encounter
with coworkers and clients. Int J Occup Med Environ Health
22:243–260
56. Nakao M, Yano E (2006) Somatic symptoms for predicting
depression: one-year follow-up study in annual health examina-
tions. Psychiatry Clin Neurosci 60:219–225
57. Pearlin LI (1999) Stress and mental health: a conceptual over-
view. In: Horwitz AV, Scheid TL (eds) A handbook for the study
of mental health-social contexts and systems. Cambridge Uni-
versity Press, New York, pp 161–175
58. Pearlin LI, Schooler C (1978) The structure of coping. J Health
Soc Behav 19:2–21
59. Plaisier I, de Bruijn JG, de Graaf R, Have Mt, Beekman AT,
Penninx BW (2007) The contribution of working conditions and
social support to the onset of depressive and anxiety disorders
among male and female employees. Soc Sci Med 64:401–410
60. Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP (2003)
Common method biases in behavioral research: a critical review
of the literature and recommended remedies. J Appl Psychol
88:879–903
61. Rasbash J, Steele F, Browne WJ, Goldstein H (2012) A user’s
guide to MLwiN. Version 2.26. Multilevel Models Project,
Institute of Education, University of London, London
62. Rau R, Morling K, Rosler U (2010) Is there a relationship
between major depression and both objectively assessed and
perceived demands and control? Work Stress 24:88–106
63. Rosenberg M (1979) Conceiving the self. Basic Books, New
York
64. Schaufeli WB, Leiter MP, Maslach C, Jackson SE (1996) The
Maslach Burnout Inventory—general survey. In: Maslach C,
Jackson SE, Leiter MP (eds) MBI manual. Consulting Psychol-
ogists Press, Palo Alto
65. Shirom A (2003) Job related burnout: a review. In: Quick JC,
Tetrick LE (eds) Handbook of occupational health psychology.
American Psychological Association, Washington, DC,
pp 245–264
66. Siegrist J (1996) Adverse health effects of high-effort/low-reward
conditions. J Occup Health Psychol 1:27–41
67. Siegrist J, Peter R (1996) Measuring effort-reward imbalance at
work: guidelines. Heinrich Heine University, Dusseldorf
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy
68. Singh P, Suar D, Leiter MP (2012) Antecedents, work-related
consequences, and buffers of job burnout among Indian software
developers. J Leadersh Organ Stud 19:83–104
69. Smelser NJ (1997) Problematics of sociology. The George Sim-
mel lecture 1995. University of California Press, Berkeley
70. Snijders TAB, Bosker RJ (1999) Multilevel analysis. In: An
introduction to basic and advanced multilevel modeling. Sage
Publications, London
71. Spence Laschinger HK, Finegan J (2008) Situational and dispo-
sitional predictors of nurse manager burnout: a time-lagged
analysis. J Nurs Manag 16:601–607
72. Stansfeld S, Candy B (2006) Psychosocial work environment and
mental health—a meta-analytic review. Scand J Work Environ
Health 32:443–462
73. Sundin O, Soares J, Grossi G, Macassa G (2011) Burnout among
foreign-born and native Swedish women: a longitudinal study.
Women Health 51:643–660
74. Tai W-T, Liu S-C (2007) An investigation of the influences of job
autonomy and neuroticism on job stressor–strain relations. Soc
Behav Personal 35:1007–1020
75. Tayfur O, Arslan M (2013) The role of lack of reciprocity,
supervisory support, workload and work-family conflict on
exhaustion: evidence from physicians. Psychol Health Med
18:564–575
76. Tepper BJ (2000) Consequences of abusive supervision. Acad
Manage J 43:178–190
77. Tepper BJ (2007) Abusive supervision in work organizations:
review, synthesis, and research Agenda. J Manag 33:261–289
78. Wang J, Adair CE, Patten SB (2006) Mental health and related
disability among workers: a population-based study. Am J Ind
Med 49:514–522
79. Wang J, Lesage A, Schmitz N, Drapeau A (2008) The relation-
ship between work stress and mental disorders in men and
women: findings from a population-based study. J Epidemiol
Commun Health 62:42–47
80. Wang JL, Schmitz N, Dewa C (2010) Socioeconomic status and
the risk of major depression: the Canadian National Population
Health Survey. J Epidemiol Commun Health 64:447–452
81. Wang Z, Inslicht SS, Metzler TJ, Henn-Haase C, McCaslin SE,
Tong H, Neylan TC, Marmar CR (2010) A prospective study of
predictors of depression symptoms in police. Psychiatry Res
175:211–216
82. Wheaton B (1994) Sampling the stress universe. In: Avison WR,
Gotlib IH (eds) Stress and mental health contemporary issues and
prospects for the future. Plenum Press, New York, pp 77–114
83. Xanthopoulou D, Bakker AB, Dollard MF, Demerouti E,
Schaufeli WB, Taris TW, Schreurs PJ (2007) When do job
demands particularly predict burnout? The moderating role of job
resources. J Manage Psychol 22:766–786
84. Xie Z, Wang A, Chen B (2011) Nurse burnout and its association
with occupational stress in a cross-sectional study in Shanghai.
J Adv Nurs 67:1537–1546
85. Ylipaavalniemi J, Kivimaki M, Elovainio M, Virtanen M, Kel-
tikangas-Jarvinen L, Vahtera J (2005) Psychosocial work char-
acteristics and incidence of newly diagnosed depression: a
prospective cohort study of three different models. Soc Sci Med
61:111–122
86. Zhong J, You J, Gan Y, Zhang Y, Lu C, Wang H (2009) Job
stress, burnout, depression symptoms, and physical health among
Chinese university teachers. Psychol Rep 105:1248–1254
Soc Psychiatry Psychiatr Epidemiol
123
Author's personal copy